nao is an open-source framework for building analytics agents, organised around context engineering: deciding what an agent needs to know about a data warehouse before it can answer a question about it. Build, evaluation and deployment sit in the same tool.
Evaluation is the part that distinguishes it. An analytics agent that is right most of the time is useless without a way to establish which answers to trust, and most frameworks stop at generation.
Being open source means it can be self-hosted and inspected, which matters when the agent has access to a company’s data. The pricing page returns a 404, self-hosting is operational work, and agents answering data questions still need their output checked against the warehouse.
